Deep Neural Networks (DNNs) have the potential to improve the quality of\nimage-based 3D reconstructions. However, the use of DNNs in the context of 3D\nreconstruction from large and high-resolution image datasets is still an open\nchallenge, due to memory and computational constraints. We propose a pipeline\nwhich takes advantage of DNNs to improve the quality of 3D reconstructions\nwhile being able to handle large and high-resolution datasets. In particular,\nwe propose a confidence prediction network explicitly tailored for Multi-View\nStereo (MVS) and we use it for both depth map outlier filtering and depth map\nrefinement within our pipeline, in order to improve the quality of the final 3D\nreconstructions. We train our confidence prediction network on (semi-)dense\nground truth depth maps from publicly available real world MVS datasets. With\nextensive experiments on popular benchmarks, we show that our overall pipeline\ncan produce state-of-the-art 3D reconstructions, both qualitatively and\nquantitatively.\n